I have an ongoing disagreement with Prof Panos about how to report a set of expired prediction markets. He claims that one should report “the historic average of similar markets”. That’s important but not urgent. First thing is to report whether one prediction market of interest has failed or succeeded. I have explained that that can’t be scientific (to Panos’ despair) —it can only be binary (the marking of “stupidity”, in Panos’ book).
The probabilistic and statistical method won’t be used when one reports about the expiry of one prediction market. People want to know how that prediction market fared. You have to report it in a binary way (right/wrong, correct/false, failure/success). Then, in a second leg, you can report “the historic average of similar markets”.
Chris:
Absolute returns are an indicator (like Nassim Taleb and Peter Schiff and Nouriel Roubini all losing money for their followers). But risk-adjusted-returns are more useful. To extend, it sounds like you care about means and prefer to ignore variances.
You should brush up on your Schrodinger–sure, a lot can be decomposed into, and modeled from binary outcomes. But ignoring the context of those outcomes is not useful. Being rejected by a hot date is a bad thing, but is actually a good thing if she is psychotic. There is so much more to it than the singular outcome of rejection. Extend to jobs, fields of study, sports and other competitive contests … and , of course, science. In fact, the science of prediction.
Cav,
Wow, philosophy to the power ten.
Are you a reader of Overcoming Bias?
I’ve added a few comments on this. The bottom line is that this market (and many like it) failed, because they were trying to predict a discrete outcome. If you miss by a small amount, you may as well be completely wrong. “Close” only counts in horseshoes, hand grenades, and now, continuous variable prediction markets.
You can read the full details at:
http://torontopm.wordpress.com/2009/06/11/why-public-prediction-markets-fail/
Paul
I forgot, I also wrote a background paper that examines the concept of discrete outcomes in prediction markets. You can find it here:
http://torontopm.wordpress.com/2009/06/11/a-lesson-in-prediction-markets-from-the-game-of-craps/
Paul